Risky-asset participation rose in the 1990s, then reversed after the mid-2000s despite continuing declines in financial technology costs. We explain this reversal by separating access costs from data costs. Lower access costs expand participation, but cheaper scalable data lets wealthy investors acquire more information, bid up prices, and reduce risk-adjusted returns for marginal, uninformed households. Using SCF and Addepar portfolio data, we document the model’s cross-sectional mechanisms: middle-wealth households exited, while informed investors earned higher returns and held riskier portfolios. Quantitatively, technological change accounts for 49% of the post-2007 participation reversal and explains the post-Covid rise.

We revisit the puzzle of limited stock market participation using qualitative methods common in other social sciences but rare in economics. Through in-depth interviews with investors and non-investors in Germany—a high-income country with low market participation—we elicit open-ended reflections on money without mentioning investing upfront. This allows beliefs and barriers to emerge naturally. We analyze these interviews using traditional human-led content analysis, complemented with a large language model (LLM)-based approach. We validate our findings using a representative survey of more than 7,000 individuals. While many known factors appear, we uncover a pervasive misconception: participation is believed to require selecting “safe” stocks, avoiding “bad” ones, and timing the market through monitoring and frequent trading. This inflates perceived costs and deters participation. Some investors overcome these barriers with support from family, friends, or trusted advisors. Notably, even active investors hold these beliefs, suggesting the misconception influences both entry and behavior in the market.

We study the effects of adult financial education using a large-scale randomized controlled trial linked to administrative credit register data. The intervention consists of a structured ten-hour personal finance course involving more than 10,000 people in Portugal. Six months after the program, treated individuals exhibit significant improvements in financial knowledge, especially among women, low-income and low-education individuals, and participants with low baseline financial literacy. The intervention also improves self-reported confidence, financial attitudes, and behaviors, including budgeting, saving, retirement planning, informed use of financial products, and financial market participation. Using loan-level administrative data, we find significant changes in debt management: treated individuals reduce reliance on high-cost credit card balances while preserving credit access, experience fewer delinquencies, and are more likely to renegotiate mortgages. Our results suggest that financial education can reduce financial literacy gaps and improve household balance sheet management, even within relatively affluent populations in high-income countries.